Inter-subject functional variability of gray and white matter in autism spectrum disorder

Autism spectrum disorder (ASD) is biologically highly heterogeneous; however, most studies focused on group-level analyses, overlooking inter-subject variability in functional connectivity (IVFC), particularly in white matter IVFC (WM-IVFC) where mechanisms and genetic influences remain unclear.

Resting-state functional magnetic resonance imaging data from 272 patients with ASD and 368 typical controls (TC) were obtained from the Autism Brain Imaging Data Exchange Project (ABIDE II) database. Gray matter IVFC (GM-IVFC) and WM-IVFC were compared between groups and correlated with symptom severity. A support vector machine (SVM) model was constructed to assess the diagnostic potential of GM-IVFC, WM-IVFC, and their combination. Transcriptome neuroimaging analyses were conducted by correlating IVFC alterations with regional gene expression data from the Allen Human Brain Atlas.

Both GM-IVFC and WM-IVFC showed regionally uneven distributions across the brain. Compared to TC, patients with ASD exhibited increased GM-IVFC mainly in the default mode and attention networks, and altered WM-IVFC mainly in the genu of the corpus callosum and superior fronto-occipital fasciculus, which were significantly associated with symptom severity. The SVM model utilizing both the GM-IVFC and WM-IVFC features yielded the best diagnostic performance (accuracy = 0.77). Transcriptome-neuroimaging associations revealed that GM-IVFC alterations were enriched in genes involved in sensory organ morphogenesis, whereas WM-IVFC alterations were linked to astrocyte-related pathways.

Our findings highlight the complementary roles of GM-IVFC and WM-IVFC, supporting their potential as biomarkers and offering novel insights into the genetic and neurobiological underpinnings of ASD.